Resistance to ROS1 inhibition mediated by EGFR pathway activation in non-small cell lung cancer

Abstract

The targeting of oncogenic 'driver' kinases with small molecule inhibitors has proven to be a highly effective therapeutic strategy in selected non-small cell lung cancer (NSCLC) patients. However, acquired resistance to targeted therapies invariably arises and is a major limitation to patient care. ROS1 fusion proteins are a recently described class of oncogenic driver, and NSCLC patients that express these fusions generally respond well to ROS1-targeted therapy. In this study, we sought to determine mechanisms of acquired resistance to ROS1 inhibition. To accomplish this, we analyzed tumor samples from a patient who initially responded to the ROS1 inhibitor crizotinib but eventually developed acquired resistance. In addition, we generated a ROS1 inhibition-resistant derivative of the initially sensitive NSCLC cell line HCC78. Previously described mechanisms of acquired resistance to tyrosine kinase inhibitors including target kinase-domain mutation, target copy number gain, epithelial-mesenchymal transition, and conversion to small cell lung cancer histology were found to not underlie resistance in the patient sample or resistant cell line. However, we did observe a switch in the control of growth and survival signaling pathways from ROS1 to EGFR in the resistant cell line. As a result of this switch, ROS1 inhibition-resistant HCC78 cells became sensitive to EGFR inhibition, an effect that was enhanced by co-treatment with a ROS1 inhibitor. Our results suggest that co-inhibition of ROS1 and EGFR may be an effective strategy to combat resistance to targeted therapy in some ROS1 fusion-positive NSCLC patients.

Figure 1. Crizotinib-resistant patient sample does not indicate ROS1 gene amplification or histologic change.

(A) Pre-treatment and post-resistance patient samples analyzed by break-apart FISH assay for ROS1. Red probes are to the 5′ region of ROS1 and green probes to the 3′ region. Values represent the mean number of signals per cell. The single 3′ signal (values underlined) is indicative of the ROS1 fusion gene copy number. (B) RT-PCR, using primers to SDC4 and ROS1 that span the fusion point, performed on pre-treatment and post-resistance tumor samples. SD2;R32 is the ‘long’ variant (fusion of SDC4 exon 2 to ROS1 exon 32) and SD2;R34 is the ‘short’ variant (fusion of SDC4 exon 2 to ROS1 exon 34). (C) Hematoxylin and eosin staining of pre-treatment and post-resistance patient samples.

Figure 2. HCC78-TR cells are resistant to ROS1 inhibition.

Cells were treated with TAE684 (A), crizotinib (B), or pemetrexed (C) as single-agents for 3 days and then analyzed by MTS assay. Values represent the mean ± SEM (n = 3–7). Calculated IC₅₀ values for TAE684: parental HCC78 = 0.14 µM, HCC78-TR = 1.09 µM, H322 = 1.42 µM, and HCC4006 = 1.15 µM. Calculated IC₅₀ values for crizotinib: parental HCC78 = 0.79 µM, HCC78-TR = 1.95 µM, H322 = 4.13 µM, and HCC4006 = 3.03 µM. Calculated IC₅₀ values for pemetrexed: parental HCC78 = 11nM and HCC78-TR = 14nM. HCC78-TR cells were significantly less sensitive than parental HCC78 cells to TAE684 (p<0.000005) and crizotinib (p<0.05) but not pemetrexed (p>0.05) as determined by student’s paired t-test.

Figure 3. HCC78-TR cells do not exhibit ROS1 gene amplification or morphological changes compared to parental HCC78 cells.

(A) Parental HCC78 and HCC78-TR cells analyzed by break-apart FISH assay for ROS1. Red probes are to the 5′ region of ROS1 and green probes to the 3′ region. Values represent the number of signals per cell. The single 3′ signal (values underlined) is indicative of the ROS1 fusion gene copy number. In the HCC78-TR line, two populations existed that differed based on the number of 5′ signals detected. (B) Representative bright field images of parental HCC78 and HCC78-TR cells.

Figure 4. Growth and survival signaling pathway activation is refractory to ROS1 inhibition in HCC78-TR cells.

Cells were treated with TAE684 (A) or crizotinib (B) for 4 hours. Lysates of the cells were then analyzed by Western blot using the indicated antibodies.

Figure 5. Growth and survival signaling pathway activation has become reliant on EGFR activity in the HCC78-TR cells.

(A) Phospho-RTK (top) and phospho-kinase (bottom) array analyses performed on untreated parental HCC78 and HCC78-TR cells. Proteins of interest are labeled. Unlabeled spots at the corners of both sets of arrays are the positive control. (B) Parental HCC78 and HCC78-TR cells were treated with TAE684, gefitinib, or a combination of both for 4 hours. Lysates of the cells were then analyzed by Western blot using the indicated antibodies. (C) Parental HCC78 (P) and HCC78-TR (T) cells were left untreated, treated with 1uM gefitinib for 4 hours, or treated with 100ng/mL EGF for 10 minutes. Lysates of the cells were then analyzed by Western blot using the indicated antibodies.

Figure 6. HCC78-TR cells have become sensitive to EGFR inhibition.

(A) Parental HCC78 and HCC78-TR cells were treated with gefitinib as a single-agent for 3 days and then analyzed by MTS assay. (B) Parental HCC78 and HCC78-TR cells were co-treated with 500nM TAE684 and gefitinib, and HCC827, H322, and H358 cells were treated with single-agent gefitinib for 3 days and then analyzed by MTS assay. Calculated IC₅₀ values: parental HCC78 (below 50% with single-agent TAE684), HCC78-TR = 0.86 µM, HCC827 = 0.04 µM, H322 = >5 µM, and H358 = 1.0uM. All values represent the mean ± SEM (n = 4).

Figure 7. EGF stimulation desensitizes parental HCC78 cells and CUTO-2 cells to ROS1 inhibition.

(A) Parental HCC78 and HCC78-TR cells were treated with TAE684 for 3 days with or without the addition of 100ng/mL EGF and then analyzed by MTS assay. Values represent the mean ± SEM (n = 3). Calculated IC₅₀ values for TAE684: parental+vehicle = 0.18 µM, parental+EGF = 0.57 µM, HCC78-TR+vehicle = 1.39 µM, and HCC78-TR+EGF = 1.45 µM. EGF significantly desensitized parental HCC78 but not HCC78-TR cells to TAE684 as determined by student’s paired t-test (p<0.05). (B) Parental HCC78 cells were treated with TAE684 for 4 hours, EGF for 10 minutes, or a combination of both. Lysates of the cells were then analyzed by Western blot using the indicated antibodies. (C) CUTO-2 cells were treated with TAE684 for 4 days with or without the addition of 100ng/mL EGF and then analyzed by MTS assay. Values represent the mean ± SEM (n = 3). Calculated IC₅₀ values for TAE684: +vehicle = 0.2 µM and +EGF = 0.81 µM. EGF significantly desensitized CUTO-2 cells to TAE684 as determined by student’s paired t-test (p<0.01). (D) CUTO-2 cells were treated with TAE684 for 4 hours, EGF for 10 minutes, or a combination of both. Lysates of the cells were then analyzed by Western blot using the indicated antibodies. Phosphorylated ROS1 bands were below the limit of detection and were therefore not included.